PointFive growth playbook

Add an "AI agents" provider to the Hub with 15 to 20 entries written in the Hub's format (the waste, the billing model behind it, how to detect it, how to fix it): cold-cache context restarts, retry loops after tool failures, frontier models on trivial tasks, oversized tool output, unused MCP servers loaded into every session, idle seats. Base each on Labs data or anonymized TokenShift metadata, link each to the matching TokenShift detection, and invite practitioners to contribute as the Hub's principles already promise.

Engineers search for the waste they see, not for a product name. The Hub already earns that search for cloud; extending it gives TokenShift a library of pages that answer coding-agent cost questions, a reference for the Creative Lead to build on, and proof of depth that gateways cannot show.

[ ] Context restarted with a cold cache
[ ] Retry loops after failed tool calls
[ ] Frontier model used for low-complexity tasks
[ ] Tool output larger than the task needs
[ ] MCP servers loaded but never called
[ ] Seats with no sessions in 30 days
[ ] Each entry: waste, billing model, detection, fix, Labs source

• Entries read as product pitches

  Keep the Hub format neutral; put the product link at the end.

• Patterns change with each agent release

  Date every entry and review it quarterly.

• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source
• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• HiringSenior Product Manager (Tel Aviv, posted October 5, 2026) to "own TokenShift, PointFive's product for finding and cutting wasted AI token spend" and "take TokenShift from early adoption to a core pillar of PointFive's platform"; the role works with "go-to-market teams on positioning, messaging, and launch plans" and tracks "adoption, cost savings, detection accuracy". The landscape to watch: "coding agents and harnesses, LLM gateways and routers, AI observability".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source

Write one page per tool in the existing compare format, framed as "with" rather than "versus" where the products complement each other: what the gateway sees (the request), what TokenShift sees (the session, repo, tools and retries), when you need both, and how they work together. Add the pages to the compare hub under a TokenShift heading.

Comparison searches come from buyers late in an evaluation. Owning the "gateway or workstation" question frames TokenShift as the missing layer rather than a competitor to tools already installed.

• Gateways ship workstation features

  Date each page and update when docs change.

• Your websiteCompare section: 18 pages, all "PointFive OS" against cloud cost tools (CloudHealth, Cloudability, Finout, Vantage, CloudZero, Kubecost, Datadog, nOps and others). None compares TokenShift. PointFive's own guide to token optimization tools names gateways (LiteLLM, Portkey, Cloudflare AI Gateway, Kong AI Gateway) and observability tools (Langfuse, Helicone) as approaches buyers combine with TokenShift.Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• HiringSenior Product Manager (Tel Aviv, posted October 5, 2026) to "own TokenShift, PointFive's product for finding and cutting wasted AI token spend" and "take TokenShift from early adoption to a core pillar of PointFive's platform"; the role works with "go-to-market teams on positioning, messaging, and launch plans" and tracks "adoption, cost savings, detection accuracy". The landscape to watch: "coding agents and harnesses, LLM gateways and routers, AI observability".Source
• Competitor researchRivals' own homepages: Finout, "FinOps Platform for the Agentic Era" that optimizes "down to the token level"; CloudZero, "The AI ROI Company"; Vantage, a "self-service cloud cost platform"; nOps, "Pay only for results". Gateways move toward spend control too: LiteLLM says it can "Track and cap LLM spend"; Portkey sells an "AI gateway, observability, guardrails"; Helicone is an "AI Gateway & LLM Observability".See the research Source

Pick one early TokenShift customer, measure 30 days of before and after (spend per active developer, cache hit rate, budget overruns avoided, policy blocks), and publish it as a short story plus a one-minute video, anonymized if needed. Use the Cloud Efficiency Architect's engagement notes as the source.

Enterprise buyers of a new category need one peer who went first. A single credible number does more for TokenShift than any feature page.

• Customer will not be named

  Publish anonymized with exact numbers, as the global software company story does.

• Your websiteCustomer stories: eight (Nubank twice, Net Health, Dolby, PwC, Blackhawk Network, Fanatics Commerce and an unnamed global software company) plus a Deloitte partner story; all are about cloud and data waste, the closest to AI being PwC "Mapping the cost of an LLM pipeline across eight AWS regions". None covers TokenShift or coding agents. The Playbook page cites "10 days to ROI" at Nubank, "$300K+ AWS savings" at Dolby and "3,000+ autonomous remediations" at Nubank, whose "Cost Champions" sit in "45+ business units".Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• HiringCloud Efficiency Architect (New York, posted September 4, 2026): "build our professional services offering from the ground up, defining what we sell, how it's packaged, and how it's delivered"; "Define best practices for AI and token spend management, positioning PointFive as the leading authority in a new and fast-evolving field"; "Drive renewals and expansion"; write, speak and attend events "in a field where the playbooks don't exist yet".Source

Build a short web tool: pick the agents, team size, typical tasks and current habits (session restarts, model choice, caching), and get an estimated monthly cost range plus the three biggest waste patterns, each linked to a Hub entry. Offer a TokenShift trial or the 30-day history view as the next step, and let people share the result card.

A tool that answers "what should this cost us" is shared inside teams, earns links, and qualifies leads by stack and team size before the first call.

• Estimates questioned

  Show the method and link the benchmark data.

• Low-intent traffic

  Ask for email only to save the result.

• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source

Package the finding as a single chart, a 60-second video from the Labs leads, a LinkedIn post from the CEO, and a short "why fewer tokens can cost more" explainer. Invite tool makers and practitioners to rerun the benchmark with their own setup and publish the results.

A counter-intuitive, reproducible number spreads on its own and makes PointFive the source on what AI work really costs, which is the Creative Lead's success measure.

• Read as attacking a vendor

  Name no tool in the headline; describe the tested setup.

• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Competitor researchRivals' own homepages: Finout, "FinOps Platform for the Agentic Era" that optimizes "down to the token level"; CloudZero, "The AI ROI Company"; Vantage, a "self-service cloud cost platform"; nOps, "Pay only for results". Gateways move toward spend control too: LiteLLM says it can "Track and cap LLM spend"; Portkey sells an "AI gateway, observability, guardrails"; Helicone is an "AI Gateway & LLM Observability".See the research Source

Add the guarantee, with its terms, to the demo page, the PointFive OS page and the FAQ, next to the customer figures (30% savings, average ROI above 1000%). State clearly whether it also applies to TokenShift.

Risk reversal is the biggest lever for a new vendor in a crowded category, and it answers the CFO question before the first call.

• Attracts poor-fit leads

  Qualify on cloud spend in the form.

• Your website"AI Casino" campaign page: four games (roulette, slots, horse race films and a card game, "Three hands, under 60 seconds", with every card "based on a detection type PointFive makes"). It is the only page found that states "Performance-based pricing. If PointFive does not save you money, you do not pay."; the homepage, demo request page, TokenShift page and FAQ do not mention it.Source
• Press releaseSeries B (June 8, 2026): $60M led by Accel with Salesforce Ventures and Index Ventures, after "6x ARR growth between 2024 and 2025"; "Customers report cloud cost savings of 30%, with an average ROI of more than 1000%". The same day PointFive launched the AI Efficiency OS (Chat, Agents, Apps) and TokenShift.Source
• Competitor researchRivals' own homepages: Finout, "FinOps Platform for the Agentic Era" that optimizes "down to the token level"; CloudZero, "The AI ROI Company"; Vantage, a "self-service cloud cost platform"; nOps, "Pay only for results". Gateways move toward spend control too: LiteLLM says it can "Track and cap LLM spend"; Portkey sells an "AI gateway, observability, guardrails"; Helicone is an "AI Gateway & LLM Observability".See the research Source

Choose one theme for re:Invent (for example "what your coding agents really cost"), and line up: the Hub's AI-agent entries, the bill check tool, the Labs finding, a TokenShift customer quote, a partner session and pre-booked meetings. Run a shared calendar and a weekly check-in until the event, then a two-week follow-up sequence.

One connected moment gets more meetings and press than four separate announcements competing in the same week.

• Too much to finish

  Cut to what is ready three weeks before.

• HiringIntegrated Marketing Lead (Tel Aviv, posted September 27, 2026): "the connective tissue of our marketing" across product marketing, content, field, PR, partners, web and agencies; "Keep one shared view of everything marketing has coming, months ahead"; "A major conference is four months out. Work backwards so that the product news, the story behind it, the press, the partners and the customer voices are all ready at the right moment"; "Three teams are planning three good ideas for the same month. Turn them into one moment."Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Your websitePartners page: "Resell it, build a services practice on top of it, or do both. No revenue minimum to apply"; billable "Efficiency assessments" and "Continuous optimization" for GSIs, FinOps consultancies and MSPs; a Capgemini partner quote; AWS Marketplace, Azure Marketplace, AWS ISV Accelerate and FinOps Foundation Premier Member.Source

Package a four-week assessment: TokenShift on a pilot team plus PointFive OS on the AI services bill, a written report on waste found, policies and budgets to set, and a savings plan. Price it as a fixed fee credited against a subscription, and let partners deliver it too.

An assessment shortens the path to a platform deal, creates the reference stories TokenShift needs, and gives the services practice a repeatable product.

• Services distract from product

  Keep the scope fixed and automate the report.

• HiringCloud Efficiency Architect (New York, posted September 4, 2026): "build our professional services offering from the ground up, defining what we sell, how it's packaged, and how it's delivered"; "Define best practices for AI and token spend management, positioning PointFive as the leading authority in a new and fast-evolving field"; "Drive renewals and expansion"; write, speak and attend events "in a field where the playbooks don't exist yet".Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• Your websitePartners page: "Resell it, build a services practice on top of it, or do both. No revenue minimum to apply"; billable "Efficiency assessments" and "Continuous optimization" for GSIs, FinOps consultancies and MSPs; a Capgemini partner quote; AWS Marketplace, Azure Marketplace, AWS ISV Accelerate and FinOps Foundation Premier Member.Source
• Your websiteCustomer stories: eight (Nubank twice, Net Health, Dolby, PwC, Blackhawk Network, Fanatics Commerce and an unnamed global software company) plus a Deloitte partner story; all are about cloud and data waste, the closest to AI being PwC "Mapping the cost of an LLM pipeline across eight AWS regions". None covers TokenShift or coding agents. The Playbook page cites "10 days to ROI" at Nubank, "$300K+ AWS savings" at Dolby and "3,000+ autonomous remediations" at Nubank, whose "Cost Champions" sit in "45+ business units".Source

Create a partner assessment kit: a Hub-based checklist by platform, a co-branded report template, a TokenShift pilot option, and a listing on the partners page for certified firms. Track assessments and the platform deals that follow.

Partners sell the method, PointFive gets the platform deal, and the Hub becomes the industry checklist rather than one vendor's blog.

• Partners skip the platform

  Make the kit depend on PointFive data.

• Your websitePartners page: "Resell it, build a services practice on top of it, or do both. No revenue minimum to apply"; billable "Efficiency assessments" and "Continuous optimization" for GSIs, FinOps consultancies and MSPs; a Capgemini partner quote; AWS Marketplace, Azure Marketplace, AWS ISV Accelerate and FinOps Foundation Premier Member.Source
• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source
• HiringCloud Efficiency Architect (New York, posted September 4, 2026): "build our professional services offering from the ground up, defining what we sell, how it's packaged, and how it's delivered"; "Define best practices for AI and token spend management, positioning PointFive as the leading authority in a new and fast-evolving field"; "Drive renewals and expansion"; write, speak and attend events "in a field where the playbooks don't exist yet".Source

For each target account, infer the stack from public signals (job posts, engineering blogs, marketplace listings) and send the three Hub entries most likely to apply, with a short video on one. For accounts rolling out coding agents, lead with the AI-agent entries and the bill check.

Specific, useful outreach earns replies from engineers who ignore vendor emails.

• Wrong stack guesses

  Phrase as "teams like yours often see".

• HiringEnterprise Sales Development Representative (Boston, hybrid): outbound "into named enterprise accounts through cold calling, email, LinkedIn", "serving as the first point of contact for inbound interest", discovery calls "with engineering leaders, FinOps practitioners, and infrastructure decision-makers", "social selling, LinkedIn engagement, and video outreach". The board also lists an Enterprise Account Executive in London. PointFive's job posts say "our go-to-market team is focused on North America".Source
• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source
• Your websiteContent: 113 blog posts and 34 guides, latest October 9, 2026, including "Cursor AI pricing 2026", "GitHub Copilot cost 2026" and "Evaluating AI coding-agent cost tools"; 964 URLs in the sitemap. The AI cost page quotes Synchrony's VP Cloud and Platform Engineering: "Token management is probably one of the most critical pieces of the overall AI landscape because that is where you tend to blow your budget the fastest."Source

Publish the MCP server in the main MCP registries and agent marketplaces with a clear README and a demo prompt list, and add the "billing data" tools when they ship. In TokenShift, show which MCP servers each team uses, with PointFive's listed as an option for cost questions.

Distribution where engineers already browse turns the MCP server into a self-serve entry point.

• Security review of the server

  Publish its permissions and data handling.

• Your websiteDevelopers / MCP page: an MCP server that brings PointFive opportunities, resources, anomalies and commitments into Claude, Cursor, ChatGPT "and any MCP-compatible AI assistant"; "Billing Data (coming soon)".Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source

Give each accepted entry a contributor byline, a shareable badge and a place on a quarterly leaderboard; run themed contribution sprints (AI agents, Kubernetes, data platforms) with the FinOps community; feature top contributors at events.

People share what carries their name. A contributor program grows the Hub and makes practitioners advocates.

• Low-quality submissions

  Keep the governance review.

• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source

Offer every PointFive OS customer a one-team TokenShift install through their MDM and a report on the previous 30 days of agent spend and waste within a week. Present it in the next business review with the platform team.

Existing customers buy faster than new ones, and a report on their own past month is hard to ignore.

• Security review delays installs

  Lead with the on-device privacy model.

• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• HiringCloud Efficiency Architect (New York, posted September 4, 2026): "build our professional services offering from the ground up, defining what we sell, how it's packaged, and how it's delivered"; "Define best practices for AI and token spend management, positioning PointFive as the leading authority in a new and fast-evolving field"; "Drive renewals and expansion"; write, speak and attend events "in a field where the playbooks don't exist yet".Source
• Your websiteCustomer stories: eight (Nubank twice, Net Health, Dolby, PwC, Blackhawk Network, Fanatics Commerce and an unnamed global software company) plus a Deloitte partner story; all are about cloud and data waste, the closest to AI being PwC "Mapping the cost of an LLM pipeline across eight AWS regions". None covers TokenShift or coding agents. The Playbook page cites "10 days to ROI" at Nubank, "$300K+ AWS savings" at Dolby and "3,000+ autonomous remediations" at Nubank, whose "Cost Champions" sit in "45+ business units".Source

Cut each recording into one-question clips (how budgets block, what the coach says, what leaves the machine), publish them on the TokenShift page, YouTube and LinkedIn, and add them to SDR sequences.

Short clips answer objections at the moment they arise and are reusable by sales.

• Outdated UI

  Re-record when the product changes.

• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source
• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• HiringEnterprise Sales Development Representative (Boston, hybrid): outbound "into named enterprise accounts through cold calling, email, LinkedIn", "serving as the first point of contact for inbound interest", discovery calls "with engineering leaders, FinOps practitioners, and infrastructure decision-makers", "social selling, LinkedIn engagement, and video outreach". The board also lists an Enterprise Account Executive in London. PointFive's job posts say "our go-to-market team is focused on North America".Source

Launch an annual "Cost Champion" award for engineers and platform teams who removed the most waste, judged on Hub categories, with a ceremony at a FinOps event and shareable winner cards. Customers nominate; non-customers can enter with their own evidence.

Awards give customers a reason to post about PointFive and give the brand a recurring moment.

• Customers decline publicity

  Allow team-level, unnamed awards.

• Your websiteCustomer stories: eight (Nubank twice, Net Health, Dolby, PwC, Blackhawk Network, Fanatics Commerce and an unnamed global software company) plus a Deloitte partner story; all are about cloud and data waste, the closest to AI being PwC "Mapping the cost of an LLM pipeline across eight AWS regions". None covers TokenShift or coding agents. The Playbook page cites "10 days to ROI" at Nubank, "$300K+ AWS savings" at Dolby and "3,000+ autonomous remediations" at Nubank, whose "Cost Champions" sit in "45+ business units".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source

With customer consent, aggregate anonymized TokenShift metadata into a quarterly report: spend per active developer, cache behavior, retry rates, model mix and the top waste patterns. Publish the method and charts, and pitch the findings to press and analysts.

Original data makes PointFive the cited source on AI cost, the Creative Lead's main goal, and gives sales a benchmark every prospect wants to compare against.

• Too few customers for stable numbers

  Start with ranges and grow the sample.

• Privacy concerns

  Aggregate only and publish the method.

• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• HiringCompany description in PointFive job posts: "We pioneered DeepWaste across cloud and data platforms, and we are now doing the same for AI token spend with TokenShift"; customers include Fanatics, H&M, Hertz, Nubank and Citizens Bank; "$96 million" raised, "most recently a $60 million Series B led by Accel"; "PointFive Labs, our research arm, publishes original work on how infrastructure and AI actually get paid for."Source

Ship a free personal edition: install on your own machine, see your own spend, cache habits and coaching tips, no central console. Offer a team upgrade when three or more people from the same domain install it.

Bottom-up adoption creates demand inside enterprises before sales arrives, and developers who see their own waste become internal champions.

• Security teams block personal installs

  Publish a security note and let admins approve.

• Your websiteTokenShift page: "Control AI agents where they run"; supports Codex, Claude Code, GitHub Copilot CLI, Cursor, Devin Desktop and OpenCode, deployed through Jamf, Intune or JumpCloud; "A gateway sees the request. TokenShift sees the session, the code and the tools behind it"; coaching example: "you restart context every session even though your cache is cold"; budgets that "warn, block, or require approval"; prompts "Never sent"; "Up to 30 days of history from day one of install."Source
• HiringSenior Product Manager (Tel Aviv, posted October 5, 2026) to "own TokenShift, PointFive's product for finding and cutting wasted AI token spend" and "take TokenShift from early adoption to a core pillar of PointFive's platform"; the role works with "go-to-market teams on positioning, messaging, and launch plans" and tracks "adoption, cost savings, detection accuracy". The landscape to watch: "coding agents and harnesses, LLM gateways and routers, AI observability".Source
• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source

Publish an open "efficiency posture" scoring framework built on Hub categories, with a self-assessment, partner-run assessments and the PointFive platform as the automated version. Invite FinOps community members to govern it.

Standards create categories. If buyers measure posture with PointFive's framework, PointFive becomes the default tool.

• Seen as vendor-owned

  Give governance to practitioners.

• Your websiteCloud Efficiency Hub (checked October 22, 2026): 531 inefficiencies and 166 service pages. Its AI entries cover cloud and API services (Amazon Bedrock, Azure OpenAI, the Claude and OpenAI APIs, Vertex AI, SageMaker, Databricks Model Serving), such as missing prompt caching and batch processing; there are no entries for AI coding agents on employee workstations (Claude Code, Cursor, Codex, Copilot). The About page promises the Hub "will expand to remain the definitive catalog of inefficiencies" as "new services and usage-based pricing models emerge", and that it is "Always open, always free" and "Community-powered".Source
• Your websitePartners page: "Resell it, build a services practice on top of it, or do both. No revenue minimum to apply"; billable "Efficiency assessments" and "Continuous optimization" for GSIs, FinOps consultancies and MSPs; a Capgemini partner quote; AWS Marketplace, Azure Marketplace, AWS ISV Accelerate and FinOps Foundation Premier Member.Source
• HiringCompany description in PointFive job posts: "We pioneered DeepWaste across cloud and data platforms, and we are now doing the same for AI token spend with TokenShift"; customers include Fanatics, H&M, Hertz, Nubank and Citizens Bank; "$96 million" raised, "most recently a $60 million Series B led by Accel"; "PointFive Labs, our research arm, publishes original work on how infrastructure and AI actually get paid for."Source

Run AMAs with the Labs leads in AI coding and FinOps communities where the rules allow, around each new paper. Answer with data, link the benchmark, and avoid product pitches.

Practitioner communities reward expertise and punish ads; an AMA gives PointFive credibility in the places buyers ask for advice.

• Hostile questions

  Answer with data and caveats.

• Your websitePointFive Labs: two papers. "Token reduction is not cost reduction" (July 2026): in the tested configuration "delivered tool-output tokens fell 38.4% while pooled billed cost rose 6.8%". "Prompt-induced waste in coding agents" (August 2026). The open AI Efficiency Benchmark publishes task manifests, retained data and analysis scripts that "can be re-analyzed without running a new paid campaign".Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source

Audit the tag container, remove unused tags, load marketing tags after the page is usable, and set a rule that new campaign pages get a speed check before launch.

Keeping the site fast protects the good field scores as campaign traffic grows, and interactive tools stay responsive on phones.

• Breaking attribution

  Test conversions after each change.

• PageSpeed testGoogle PageSpeed for the pointfive.co homepage (October 22, 2026): real Chrome visitors pass Core Web Vitals (main content in 1.4 s, taps respond in 112 ms, no layout shift). The mobile lab test scores 68/100 with Largest Contentful Paint 9.8 s; Google Tag Manager is 516 KB with 428 ms of blocking, and about 248 KB of JavaScript is unused.Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source

Run the card game on tablets at the booth with a live leaderboard, a daily prize, and a short follow-up email showing the detections the player picked and how often PointFive finds them.

A game draws a crowd, qualifies players by the cards they choose, and gives SDRs a personal follow-up.

• Play without interest

  Score leads by job title.

• Your website"AI Casino" campaign page: four games (roulette, slots, horse race films and a card game, "Three hands, under 60 seconds", with every card "based on a detection type PointFive makes"). It is the only page found that states "Performance-based pricing. If PointFive does not save you money, you do not pay."; the homepage, demo request page, TokenShift page and FAQ do not mention it.Source
• HiringCreative Lead (New York, posted October 2, 2026; a second post in Tel Aviv, September 27): "The goal is buzz, shares, followers, mindshare and category authority. Pipeline follows attention, but attention is the job." Briefs include "Turn PointFive's research into stories that engineers, executives and press repeat", "Build something interactive, a tool, a game or a quiz, that people send to their team" and "Our category all looks and sounds the same." Success: "PointFive becomes the name people bring up first when the topic is infrastructure cost and AI cost." It is the first such hire: "no playbook, no asset library and no creative team underneath you."Source
• Your websiteEvents page (October 22, 2026): upcoming AGNTCon + MCPCon (October 22-23) and AWS re:Invent (Las Vegas, November 30 - December 4); past FinOps X San Diego and London, Snowflake Summit, Databricks Data + AI Summit, Google Cloud Next, a Denver FinOps meetup and an Amsterdam reception; on-demand webinars include two TokenShift demos (July 22 and August 19).Source